With Anthropic and the US Department of Defense still in the process of mending their relationship, there remains a vacuum for other companies to pick up the slack and offer AI services for dubious purposes. Google employees are making it clear to their employer that they have no interest in filling that gap. More than 600 Googlers, including members of the company’s DeepMind lab, signed a letter to CEO Sundar Pichai demanding that he refuse to allow Google’s artificial intelligence tools to be used by the Pentagon for classified work.
The signees—which, according to The Verge, included principals, directors, and vice presidents within the company—noted that their work on AI has made them acutely aware of the potential dangers of using the technology without guardrails. Given that the Pentagon demanded Anthropic ignore its red lines for AI use and allow the agency to do anything it wants, including performing domestic surveillance and using fully autonomous weapons, it’s a reasonable concern for Googlers to raise.
“As people working on AI, we know that these systems can centralize power and that they do make mistakes,” the employees wrote in the letter. “We feel that our proximity to this technology creates a responsibility to highlight and prevent its most unethical and dangerous uses.”
To that end, they explicitly mentioned the very same clauses that created issues for Anthropic as their primary objections. “We want to see AI benefit humanity; not to see it being used in inhumane or extremely harmful ways. This includes lethal autonomous weapons and mass surveillance but extends beyond,” the letter reads.
The employees argued, “The only way to guarantee that Google does not become associated with such harms is to reject any classified workloads. Otherwise, such uses may occur without our knowledge or the power to stop them.” Google has yet to publicly acknowledge the letter, but it seems the company has been playing footsie with the military. A report from The Information earlier this month suggested that Google was exploring a deal with the Pentagon, similar to the one OpenAI entered into when it offered up its services following the DoD’s spat with Anthropic.
The fact that OpenAI and Google have decided, as corporations, to do business with the Pentagon, which has demanded fealty from any AI partners so they can use the models for “all lawful uses,” is pretty cowardly. Especially considering both firms filed amicus briefs in support of Anthropic in its lawsuit against the government after it was labeled a “supply chain risk.” They’re basically letting Anthropic take the hit of actually fighting the fight against the Pentagon while raking in lucrative contracts with the agency. They’re also doing so in direct opposition to their respective employees, as OpenAI workers also petitioned leadership not to do business with the military without specific restrictions in place.
Read the full article here
